@inproceedings{wang-etal-2026-solar,
    title = "{SOLAR}-{RL}: Semi-Online Long-horizon Assignment Reinforcement Learning",
    author = "Wang, Jichao  and
      Bian, Liuyang  and
      Zhou, Yufeng  and
      Xiao, Han  and
      Pan, Yue  and
      Wang, Guozhi  and
      Wang, Hao  and
      Wang, Zhaoxiong  and
      Wen, Yafei  and
      Chen, Xiaoxin  and
      Ren, Shuai  and
      Zeng, Lingfang",
    editor = "Liakata, Maria  and
      Moreira, Viviane P.  and
      Zhang, Jiajun  and
      Jurgens, David",
    booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {ACL} 2026",
    month = jul,
    year = "2026",
    address = "San Diego, California, United States",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingest-acl/2026.findings-acl.1222/",
    pages = "24419--24432",
    ISBN = "979-8-89176-395-1",
    abstract = "As Multimodal Large Language Models (MLLMs) mature, GUI agents are evolving from static interactions to complex navigation. While Reinforcement Learning (RL) has emerged as a promising paradigm for training MLLM agents on dynamic GUI tasks, its effective application faces a dilemma.Standard Offline RL often relies on static step-level data, neglecting global trajectory semantics such as task completion and execution quality. Conversely, Online RL captures the long-term dynamics but suffers from high interaction costs and potential environmental instability. To bridge this gap, we propose SOLAR-RL (Semi Online Long-horizon RL). Instead of relying solely on expensive online interactions, our framework integrates global trajectory insights directly into the offline learning process. Specifically, we reconstruct diverse rollout candidates from static data, detect the first failure point using per-step validity signals, and retroactively assign dense step-level rewards with target-aligned shaping to reflect trajectory-level execution quality{---}effectively simulating online feedback without interaction costs.Extensive experiments demonstrate that SOLAR-RL significantly improves long-horizon task completion rates and robustness compared to strong baselines, offering a sample-efficient solution for autonomous GUI navigation."
}